Baserow ORM

The Baserow Python client comes with basic ORM capabilities.

Note: The ORM API is not Mypy compatible. Support could be added by implementing a Mypy plugin.

Defining models

# myapp/models.py

from baserow.orm import Column, ForeignKey, Model

class Product(Model):
  name = Column('Name')
  price = Column('Price')

class Customer(Model):
  name = Column('Name')
  favorite_products = ForeignKey('Favorite Products', Product)

Generating a database mapping

Because the database schema cannot be introspected using a normal token, it is necessary to generate a mapping for the ORM using a JWT. The easiest way to do this is to use the Baserow ORM command-line interface.

$ python -m baserow.orm \
    'My web shop' myapp.models.Product:Produces myapp.models.Customer:Customers \
    --url https://baserow.io --user my-email@example.org --write-to var/conf/mapping.json

You can specify --password '...' to avoid the password prompt, or directly pass a JWT with --jwt '...'.

Database connection

The database connection must be initialized with a Baserow client and the database mapping that was generated in the previous step. Since the ORM can only perform CRUD operations, a long-lived token can (or should) be used.

# myapp/__main__.py

from baserow.client import BaserowClient
from baserow.orm import Database, DatabaseMapping
from .models import Product, Customer

client = BaserowClient('https://baserow.io', token='...')
db = Database(client, DatabaseMapping.load('var/data/mapping.json'))

Querying

When querying rows from Baserow with the ORM interface, you are working with instances of the Model subclasses that you have defined previously. Linked rows are queried lazily (i.e. iterating over Customer.favorite_products will fetch each linked Product from Baserow).

query = db.select(Customer).filter(Customer.name.contains('Alice'))

print('Alice likes:')
for product in query.first().favorite_products:
  print(f'- {product.name}')

Note: Fetching linked rows currently happens individually and can thus be rather slow. More information at baserow#601. You can still access the raw (id, name) pairs returned by the Baserow API for the ori

The Database.save() currently provides very naive implementation to save new or update existing rows. It does not currently handle single_select and link_row fields properly.

row = Product(id=None, name='Soy beans', price=1.99)
db.save(row)
print(row.id)